化學品用量管理最適化參數研究-以某半導體廠為例

Optimization for control parameters of chemicals dosing in a semiconductor fabrication plant

彭冠霖(Kuan-Lin Peng)

指導教授 ： 黃志彬

國立交通大學/工學院/工學院永續環境科技學程/碩士(2018年)

摘要

半導體晶圓廠製程程序複雜多樣，製程機台常需使用各式各樣的化學品，其中用量大的化學品由廠務中央化學供應系統供應給各機台使用。當機台異常造成化學品使用量增加，無法即時發現將造成化學品的浪費，增加公司的成本與風險。本研究將原有化學品管理系統增加控制參數，以利發現生產機台異常，而造成少量多次使用化學品狀況，並進行管理，以降低製程機台異常時化學品的消耗量。
研究結果顯示，既有管理系統可以增加「日需求時間」與「日需求次數」兩種參數，並且在各機台需求狀況分開整理分析後，獲得控制數據，作為最適化參數與設定值的重要參考，再依據自動用量分析結果，由管理者調整參數設定。經6個月的自動用量分析功能驗證過程，發現新增的日需求時間參數皆能發報異常通知，發現化學品異常使用狀況的達成率100%；而新增的日需求次數參數就無法在每次異常時發報異常通知，發現化學品異常的達成率僅62.5%。是故，日需求時間比日需求次數更適合作為化學品管理系統的管理參數，可以有效改善機台少量多次異常使用化學品時無法即早發現的問題。唯機台的使用方式比原本預設的複雜，導致由警報發現到的實際異常正確雖僅為18.5%，但異常狀況都有被攔截到，仍是令人滿意的。

關鍵字

化學品使用管理 ； 化學供應系統 ； 日需求時間

並列摘要

Manufacturing process of modern semiconductor devices are complex as well as use large amount of various chemicals, which are provided by the central chemical supply system.
As the manufacturing machines run abnormally, the amount of chemicals fed by the central factory increases. If this abnormality fails to be found at the earliest, the operational cost and risk will increase. By adding new parameters to the current chemicals management system, we might detect the machine failure immediately. This would lead to a higher efficiency of the chemicals management system as well as avoidance of unnecessary dose of chemicals.
The results have shown that by adding the daily demand time (DDT) and daily demand frequency (DDF) to current chemicals management system, the system is able to detect the abnormal situations of the machines. These two parameters were obtained by analyzing the demand status of each machine. From these analyses, the reference values of these two parameters were also determined. After 6-month testing, the DDT can be an indicator to detect the abnormality, and the probability of detecting the abnormal dose of chemicals is 100%; however, the detective probability of the abnormal dose of chemicals is 62.5% with the DDF. Therefore, the DDT is a more appropriate indicator than the DDF in detecting the abnormal dosing. Therefore, the DDT is an effective control parameter for the current management system. Although the probabilities of detecting the abnormal dosing were not always 100%, all abnormal situations were reported. Overall speaking, using DDT as an indicator to monitor chemical dosing in the manufacturing process of semiconductor fabrication plant is an effective approach.

並列關鍵字

chemical usage management ； chemical supply system ； daily demand time

參考文獻

鍾尚仁，半導體晶圓廠化學供應區酸性氣體污染改善研究，國立交通大學工學院永續環境科技學程，2017年。

三菱化學供應系統技術手冊，1995年。

力晶半導體新人教育訓練講義，2004年。

廖建凱，Vidas系統技術手冊，煒晶科技，2014年。

納谷嘉信，應用新QC七大手法，和昌書局，2001年。